ECOMORPHOLOGICAL ASSOCIATIONS OF FEEDING HABITS IN HYPERCARNIVORES By Sheridan Teague Kelley A THESIS Submitted to Michigan State University In partial fulfillment of the requirements For the degree of MASTER OF SCIENCE Zoology 2011 ABSTRACT ECOMORPHOLOGICAL ASSOCIATIONS OF FEEDING HABITS IN HYPERCARNIVORES By Sheridan Teague Kelley The classification scheme for Carnivora put forth by Van Valkenburgh (2007) is frequently used as a guideline for feeding ecology throughout the history of the Order. However, the categorization system used is too broad and may lead to species being classified into incorrect ecomorphs. Further, many ecomorphological studies rely on linear measurements, which fail to capture the full change in skull shape across species. We analyzed 7 hypercarnivorous species using geometric morphometric techniques to assess its effectiveness in conjunction with ecomorphological analyses, and partial least squares analysis to test the adequacy of Van Valkenburgh’s classification system. We found a surprising amount of shape variation throughout the wolf-like hypercarnivores, with some species appearing more similar to those within the other categories. Morphological differences were also observed among the bone eating hyenids. Our analyses suggest that Van Valkenburgh’s hypercarnivore classifications are in need of revision. We suggest a system that subdivides the wolf-like ecomorph into several, more distinct categories. ACKNOWLEDGEMENTS I’d like to thank my graduate committee (Barbara Lundrigan, Kay Holekamp, Tom Getty, and Meredith Gore) for their suggestions and support throughout the entirety of this project. Thanks for Lauren Phillips, Jeremy Phan, Gwen Webster, and the 2008-’09 students of “200H: Research for Undergraduates” for their assistance in photographing and digitizing the specimens. Miriam Zelditch, David Sheets, Suzanne LaCroix, and Jaime Tanner were all a great help in furthering my understanding of geometric morphometrics, both in practice and in association with the species included within this study. Thanks to Eli Swanson for his suggestions and feedback regarding the phylogenetic aspects of my research. Finally, I’d like to express my appreciation to my wonderful wife, my family, and my friends for all of their encouragement, support, and boundless patience. iii TABLE OF CONTENTS LIST OF TABLES…………………………………………v LIST OF FIGURES………………………………………..vii INTRODUCTION…………………………………………1 METHODS…………………………………………………5 RESULTS…………………………………………………..12 DISCUSSION………………………………………………17 APPENDICES……………………………………………..41 List of included museum specimens……………..…43 Locations of landmarks and semi-landmarks……....56 WORK CITED……………………………………………58 iv LIST OF TABLES Table 1 - Chart from Van Valkenburgh 2007 showing the occurrence of ecomorphs throughout the evolution of the carnivoran families, with emphasis on terrestrial species larger than 7kg in mass. A “+” indicates that at least one species in the fossil record is known to have displayed a particular ecomorph. “?” indicates that the existence of a species with a particular ecomorph is unknown given the limited fossil record. ……………..24 Table 2 – Descriptions of the common behavioral traits assigned to prey acquisition and consumption for each species used in the study following a literature review. Characteristics assigned to species are taken as common tendencies and do not reflect behaviors that occur infrequently. ……………..25 Table 3 – Index of ecological categories and the values assigned to them for use in partial least squares analysis. Values for ecomorph were ordered arbitrarily. ……………..26 Table 4 – Values assigned to each species following the ecological index in Table 3. Assignment of values was performed by comparing the descriptive summaries from Table 2 to the categories described in Table 3. ……………..27 Table 5 – Results from the Goodall’s F test with permutation for signs of sexual dimorphism in shape, and the subsequent Mancova for shape with size as a covariate that were significant prior to Bonferroni adjustment (p<0.002). ……………..28 Table 6 – Canonical variates (CVs) obtained for all 3 views with the percent of variance explained by each CV and cumulative percentage of explained variance for that view. ……………..29 Table 7 – Results from partial least squares analyses examining covariation between skull morphology and ecological variables (Table 4). The RV coefficient for a given view is inserted below the name of the corresponding view in the first column. ……………..30 v Table 8 – Weighting coefficients assigned to each ecological variable for the first 2 pairs of partial least square axes for each of the 3 views following partial least squares analysis. ……………..31 Table 9 – Blomberg’s (2003) K statistic and the p-values for tests of phylogenetic independent contrast variance under each of the 4 branch length estimation methods tested for all ecological variables, and for the consensus tree obtained from the 4 single trees. 1 = all branch lengths set to 1.0, G = Grafen (1989), N = Nee (cited in Purvis 1995), and P = Pagel (1992). P-values <0.05 show signs of non-random phylogenetic signal. ……………..32 Table 10 - List of specimens used in this study. Abbreviations within the Museum column represent where the specimen is located. (MSU = Michigan State University, East Landing, MI, TAU = Tel Aviv University, London, England, BMNH = British Museum of Natural History London, England, FMNH = Field Museum of Natural History, Chicago IL, NMK = National Museums of Kenya, Africa, RCSOM = Royal College of Surgeons, London, England). Specimens whose sex could not be determined were marked “?” in the Sex column. Lateral, ventral, and mandible refer to the digitization views used. Those specimens marked "Y" were included in the corresponding view analyses, whereas that marked with a "-" were not. ……………..42 Table 11 - Locations of landmarks and semi-landmarks for each of the three views. ……………..55 vi LIST OF FIGURES Figure 1 - - Lateral, mandibular, and ventral views of a Crocuta crocuta skull, showing locations of landmarks (red, numbered dots, see Table 2), and semi-landmarks (red, unnumbered triangles). For interpretation of the references to color in this and all other figures, the reader is referred to the electronic version of this thesis. ……………..33 Figure 2 - Phylogenetic tree of the 7 Carnivora taxa included in this study. Relationships are based on Eizirik et al. 2010. ……………..34 Figure 3 - Plot of lateral view CV1 vs. CV2 for species, with deformation grids depicting shape change along CV1 (bottom right) and CV2 (top left). The legend (bottom left) lists the symbol associated with each species. ……………..35 Figure 4 - Plot of mandibular view CV1 vs. CV2 for species, along with its deformation grids and legend as described in Figure 3. ……………..36 Figure 5 - Plot of ventral view CV1 vs. CV2 for species, along with its deformation grids and legend as described in Figure 3. Note the axis ranges are significantly less than that seen in the other 2 views. ……………..37 Figure 6 – Plot of partial least squares for shape axis 1 vs. ecology axis 1 for the lateral view. A plot of PLS1 vs. PLS2 for the ecological traits showing the directionality and relative strength of the trait coefficients is inserted to the right of the plot. Below the plot is a deformation grid showing the shape change associated with left to right movement along shape PLS1. ……………..38 Figure 7 – Plot of partial least squares for shape axis 1 vs. ecology axis 1 for the mandibular view. See Figure 6 for further explanation. ……………..39 vii Figure 8 – Plot of partial least squares for shape axis 1 vs. ecology axis 1 for the ventral view. See Figure 6 for further explanation. ……………..40 viii INTRODUCTION The Order Carnivora includes only 286 recognized species today (Eizirik et al. 2010), but its members have expanded to cover a vast array of ecological niches both in the past and the present. Carnivora is of particular interest because of its diversity in both form and function, with several common adaptive types appearing independently in more than one lineage. Van Valkenburgh (2007) attributed this repeated development of similar forms to two factors, the first being that there are a limited number of ways to partition the carnivore niche ecologically, and the second, that there have been no changes in the basic properties of muscle, skin, and bone throughout the history of Carnivora. If these two factors act as constraints, then evolution will lead to similar forms developing throughout time, because such forms would succeed in meeting similar functional demands, such as the need to break apart hard foods without damaging the feeding apparatus. Many studies have focused on dentition and linear measurements of the jaw in order to compare species with respect to morphology and associated dietary habits. Recently, there has been increasing interest in gross morphological shape and its correlation with diet. While linear measurements and area estimates can provide useful information such as bite force and feeding gape (Christiansen and Adolfssen 2005, Christiansen and Wroe 2007, Wroe et al. 2005), studies of morphological shape as a whole provide a more complete picture of the associations among the various modules that make up the feeding apparatus. Goswami and Polly (2010) suggested that trait integration for cranial modules may limit the morphological variation possible for some traits. Thus, analyses using gross morphological shape can offer greater insight than simple, linear measurements into potential selective forces and constraints that might influence both the morphology and ecological niche of a species. 1 Ecomorphology.— By looking at morphological characteristics and the ecological habits of the organisms that possess them, scientists can make inferences about the ecological relevance of different morphologies. Such inferences can informate about of the mechanical requirements associated with a particular diet (e.g., soft or hard foods), and of the evolutionary history of certain morphological traits associated with dietary habits (e.g., the degree of undulation in the banding pattern of tooth enamel seen in Hyaenidae and its relationship to bone eating behavior, as described by Ferretti (2007)). Raia (2004), for example, used geometric morphometrics to compare mandibular shape in 18 extant and 3 extinct species from the Order Carnivora. He found that mandibular shape correlates strongly with diet in large carnivores, although phylogeny accounted for the greatest portion of shape variation. Paleontological research makes wide use of ecomorphology by comparing morphological characteristics of living organisms with known ecological and dietary habits to extinct taxa in order to make ecological inferences about extinct species. In one such study, Werdelin and Solounias (1996) devised a categorization method for Hyaenidae in which they divided the various taxa among six ecomorph types based on morphological form and function, as well as expected dietary habits. Their categorization method has allowed scientists to trace the evolution of the bone breaking morphology seen in extant hyenas throughout the known history of the Family, and can be used as a starting point in identifying potential familial relationships among newly discovered species. Such categorization methods can be informative at higher levels of taxonomy as well. A review by Van Valkenburgh (2007) summarized the array of ecomorphs seen among large 2 (>7kg), terrestrial members of the Order Carnivora in both extinct and extant taxa (Table 1). She described three major ecomorphs based on predatory method, dietary composition, and phylogenetic placement: hypocarnivores, in which the diet is >70% non-vertebrate foods; mesocarnivores, with a diet of 50-70% meat and the remainder non-vertebrate foods; and hypercarnivores, defined as having a diet of >70% vertebrate prey. Hypercarnivores were further divided into three generalized ecomorphs: cat-like (including sabertooth cats and those with conical canine teeth), wolf-like, and hyena-like. Van Valkenburgh noted that the generalized mesocarnivorous morphology was common in the early members of most families of Carnivora, and that this morphology provides the starting point for species to evolve towards or away from a diet focused on vertebrate food. The development of a feeding morphology specialized for a particular diet is often accompanied by a loss of features that are not relevant to that diet (e.g., the reduction in cheek teeth in the insectivorous Aardwolf), making it unlikely that a radically different specialization will develop in the descendents of such a species (Holliday and Steppan, 2004). Species within a given ecomorph tend to have similar adaptations in dentition and skull shape. For instance, postcarnassial grinding dentition is greatly reduced or absent in all three hypercarnivorous forms, and cranial and jaw characteristics that increase the mechanical advantage of the jaw adductors (temporalis and masseter) are shared among hypercarnivorous ecomorphs. Differences among hypercarnivorous taxa are thought to reflect differences in prey acquisition (e.g., whether they chase and drag down their prey or pounce and deliver a killing bite) and dietary preferences (e.g., whether bone is a significant portion of the diet). The diversity of these ecomorphs within families, and the repeated evolution of similar forms across families, suggest that certain combinations of morphological characteristics are matched to 3 particular ecological niches, and that species with a morphology that allows for adaptation to those niches may fill a vacant niche when such an opportunity arises. Although it is clear that certain morphological characteristics are vital to the acquisition and processing of certain types of foods, the extent to which the skull as a whole varies in response to dietary composition is not well defined. The skull is made up of several parts that serve various functions (e.g., vision and the orbits, hearing and the auditory bullae), and their associated functional requirements and morphological constraints leads to a lack of developmental independence in those parts. Studies such as that by Sacco and Van Valkenburgh (2004) have developed indices of linear measurements that reflect dietary habits. However, each measurement only takes into account a small portion of the variation in shape seen across different dietary ecologies, and a particular morphological characteristic is not likely to be independent of those that are used in conjunction with it. Linear measurement techniques can also be time consuming, with only a fraction of the measurements taken resulting in significant association with dietary habits. Analysis of the overall differences in the shape of the feeding apparatus between species with different dietary habits is required for a more thorough understanding of the morphological characteristics associated with a given feeding ecology. Such studies are of great value in furthering our understanding of trait evolution and the ecology of extinct taxa. Aims of this Research.— In this study, I use geometric morphometrics to investigate the skull morphology of members of the hypercarnivorous ecomorphs described by Van Valkenburgh (2007). I focus 4 specifically on 7 species, representing 4 extant families, to assess the variation seen in morphology of the cranium and jaw and examine their associations with dietary habits and hunting strategies. My primary goal is to identify correlations between skull shape, food acquisition, and diet, and to evaluate the adequacy of the classification system as it now stands. Van Valkenburgh’s classification system is a useful tool in the determination of ecomorphological associations in extant and extinct members of Carnivora, but I believe the hypercarnivore sub-categories are too broad and in need of revision so that future ecomorphological studies can more easily and accurately classify ecological habits of extinct species. Two Block Partial Least Squares analysis is used to examine the relationship between skull morphology, visualized as relative warps that define the changes in the relative positions of pre-determined landmarks from one morphology to the other, and feeding ecology, defined using a categorical rating system that takes into account food acquisition and consumption habits. Phylogenetic relatedness is then assessed in order to determine the extent to which phylogeny might explain the observed patterns of morphological and ecological variation. METHODS Specimens.— Craniums and mandibles of 419 individuals, representing 7 extant species and 4 families from the Order Carnivora were borrowed from the following museum collections: Michigan State University Museum (East Lansing, Michigan), University of Michigan Museum of Zoology (Ann Arbor, Michigan), Tel Aviv National Museum of Natural History (Tel Aviv, Israel), British Museum of Natural History (London, England), Field Museum of Natural History 5 (Chicago, Illinois), the National Museum of Kenya (Africa), and the Royal College of Surgeons (London, England). All specimens were adults, based on the presence of a fully erupted adult dentition. Some outliers (described below) were removed after preliminary analyses because the cranium or jaw showed signs of significant deformation due to disease, improper specimen preparation, or improper specimen care. Specimens included in analyses are listed in Appendix 1 along with their museum catalog number, sex, and the views that were used in geometric morphometric analyses. Digitization.— Skulls were photographed using a Fuji FinePix S5 Pro digital camera with a 70mm Sigma macro lens. Craniums were photographed in the ventral view with the palate parallel to the image plane, and in the lateral view with the anteroposterior axis parallel to the image plane. Intact mandibles were photographed in lateral view with the mid-sagittal axis parallel to the image plane. A scale was included in all photographs. Landmarks, visibly observable and biologically meaningful locations that are persistent across all taxa used in the study, were digitized onto images of the skulls in all three views to help define the shape of the specimens. Twenty-seven landmarks were selected from the ventral cranium for analysis of shape in that view (Figure 1, Appendix 2). Landmarks alone cannot fully capture the complexity of the dorsal curve of the lateral cranium or the mandibular ramus, so these were analyzed using a combination of landmarks and semi-landmarks. Fourteen landmarks and 32 semi-landmarks were selected from the lateral view (Figure 1, Appendix 2), and 11 landmarks and 75 semilandmarks were chosen from the mandibular view (Figure 1, Appendix 2). Landmarks and semi- 6 landmarks were digitized using tpsDig2.15 (Rohlf 2010). Semi-landmarks were obtained by applying the curve-tracing function and selecting “resample,” which results in even spacing between points. Morphometric Analyses.— Landmarks were superimposed using Generalized least squares (GLS) Procrustes Analysis, a procedure that removes variation in scale, position, and orientation (Zelditch et al. 2004). Bilaterally homologous landmarks in the ventral view were first reflected and averaged in order to reduce the error that such estimations may cause. Semi-landmarks require a specialized superimposition method because their spacing is biologically arbitrary. In this study, semilandmarks were superimposed so as to minimize the Procrustes distance from the mean shape. Skull size was quantified using centroid size, which is defined as the square root of the summed square distances from each landmark to the geometric center of an object. Landmark superimposition was performed using Coordgen6f (Sheets 2009). Semi-landmarks were superimposed using Semiland6 (Sheets 2003). Reflection and averaging of bilaterally homologous landmarks was done in Sage (Marquez 2007). To identify outliers, principal components analyses were performed using PCAGen6 (Sheets 2001) with specimens grouped by species. Extreme outliers were re-examined and eliminated if it appeared that they were the result of significant damage caused by disease, injury, or specimen breakage. Sexual Dimorphism.— 7 Species were measured individually for evidence of sexual dimorphism prior to all further analyses to determine whether males and females should be combined or treated separately in subsequent analyses. Sexual dimorphism in skull shape was examined in each species individually using Goodall’s F test with 2,500 permutations for each of the 3 views. Cases of significant sexual dimorphism (p-value <0.05) were further analyzed using permutation tests on explained variance with centroid size as the covariate in order to determine the percentage of shape variation explained by differences in size, and the percentage explained by sex. A Bonferroni adjustment (which corrected the p-value for significance to p<0.002) was used to account for the multiple tests associated with each species. Goodall’s F test was performed in Twogroup6h (Sheets 2003), and the permutation test on explained variance was performed in Manovaboard6 (Sheets 2006). Ecological Index.— Ecological factors associated with food acquisition and diet were identified for use in ecomorphological correlation assessment, and the hunting and feeding ecologies of the species were classified following a review of the literature (Table 2). Species were categorized based on the importance of meat and bone in their diet, the size of their hunting groups, and the methods typically used to kill their prey. A numerical index (Table 3) was created for each category for use in partial least squares analysis (see below). The value associated with each option within a category does not express any degree of significance or strength, but rather acts as a marker for grouping species with similar ecological behaviors together for use in partial least squares analysis. For the meat and bone categories, consumption was scored (1-3, or 1-4, respectively), such that higher values correspond to greater dietary importance. The hunting party category 8 was scored (1-4) based on group size, with 1 representing solitary hunters and foragers, 2 species that infrequently form small groups based on food availability, 3 species that frequently hunt in small groups, and 4 representing obligate pack hunters. Killing strategy ranged from 1 to 5, with 1 representing species that primarily scavenge rather than hunt. Mid-range values (2 and 3) correspond with chasing strategies, with 2 involving knocking down the prey and 3 a biting or worrying strategy. Higher values (4 and 5) were associated with species that stalk their prey. Species rated as 4 perform a killing blow, while a species rated as a 5 kills by brute force or crushing. Table 4 summarizes the values assigned in each category for all species. Preliminary Analysis and Ordination.— Species were compared to one another in all 3 views using Goodall’s F test with 2,500 permutations in Twogroup6h (Sheets 2003) to evaluate whether the differences in shape between them were significant. P-values <0.02 (following Bonferroni adjustment) for species comparisons could be indicative of significant shape differences and suggest that the species can be reliably distinguished from one another. Canonical variates analysis (CVA) was used (CVAGen6, Sheets 2003) to simplify the descriptions of differences between species. This method rescales the axes of the new coordinate system so as to maximize the ability to discriminate between groups (Zelditch et al. 2004). Bartlett’s test for differences in the value of Wilk’s lambda (Λ), which is the within-groups sum of squares divided by the total sum of squares, was used to determine how many CVs were effective at discriminating between species. An assignment test was performed in CVAGen6 in which the Mahalanobis’ distance between group means was used to determine the probability that each specimen is indeed a member of the 9 group to which it is assigned (Zelditch et al. 2004), and a jack-knife groupings test was performed to obtain the rate at which specimens were correctly classified. Visual representations of the morphological differences between groups described by CVs 1 and 2 were obtained in the form of CVA deformation grids. Skulls from each species were visually compared to identify additional features not found through the morphometric analyses that might be relevant to the ecological variables of interest. The percentage of variance explained by the CVs for species in each view were obtained using MorphoJ 1.02j (Klingenberg 2011). Partial Least Squares. — Two Block partial least squares (PLS) analysis was used to investigate the correlations between morphological features and ecological strategy, with each of the three views examined separately. In each PLS analysis, the first block consisted of the Procrustes coordinates for all individuals of all species in a given view. The second block contained the ecological index scores for all specimens, with all individuals within a species having the same index scores. Partial least squares predictor coefficients were obtained for the ecological index scores along with the RV coefficient, which is a multivariate analogue of the squared correlation coefficient, for the overall strength of association between the blocks. A permutation test was performed (10,000 iterations) to test the data against the null hypothesis of complete independence between the two blocks of data. Singular values and pairwise correlations were obtained for each PLS axis pair, along with the percent of covariance explained by each axis pair. Partial least squares analysis was performed in MorphoJ 1.02j (Klingenberg 2011). 10 Phylogenetic Signal.— Similarities in shape among taxa due to phylogenetic relatedness can lead to errors in interpretation of morphometric data by obscuring the degree to which species have come to differ through evolutionary processes. A species is generally expected to be more similar in form and function to those more closely related to it than to those from a distantly related clade. Analyses were thus performed to measure the extent to which morphological differences among species are explained by phylogeny. Phylogenetic trees for the 7 species of Carnivora were constructed following the phylogeny put forth by Eizirik et al. (2010) (Figure 2). Arbitrary branch length methods were used in place of estimated branch lengths due to uncertainty in the divergence times among some of the taxa. Four trees, differing only in branch length, were generated through pdtree (Modford et al., 2010) in Mesquite for comparison of phylogenetic signal across various arbitrary branch length methods, with 1 tree each following the methods put forth by Grafen (1989), Nee (cited in Purvis 1995), and Pagel (1992). The fourth tree had all branch lengths set to 1.0. The trees were exported from Mesquite as separate NEXUS files for use in statistical analyses. A consensus tree was then generated from the 4 branch length method trees. The NEXUS files for the trees were created using Mesquite 2.74 (Maddison and Maddison, 2010). The consensus tree was used to map the Procrustes coordinates of each view to the phylogeny in MorphoJ 1.02j (Klingenberg 2011). A permutation test (10,000 iterations) was performed against the null hypothesis of no phylogenetic signal. Views found to have a p-value <0.05 have significant evidence of phylogenetic signal in skull shape. All phylogenetic trees were imported into R 2.12.1 (R Development Core Team 2011) along with the matrix of 11 ecological index ratings for statistical testing of phylogenetic signal in the ecological variables. Values for Blomberg's K (Blomberg 2003) were obtained through picante (Kembel et al. 2010) in R. Blomberg’s K is a measure of phylogenetic signal that compares the observed signal of a trait (i.e., ecological strategies) to an expected signal obtained under the Brownian motion model of trait evolution. A K value close to zero suggests random evolution or convergence, a value of 1 indicates some conservatism or phylogenetic signal, and values greater than 1 represent strong phylogenetic signal. The ecological traits were tested with each of the four phylogenetic trees and the consensus tree, and the K values obtained under each arbitrary branch length method were compared in order to determine the influence of phylogenetic signal on ecological strategy. P-values, representing the quantile for the observed phylogenetic independent contrast variance versus the null distribution, as described by Kembel (2010), were also obtained. Traits with a pvalue <0.05 have non-random phylogenetic signal. RESULTS Sexual Dimorphism.— The polar bear, African lion, and gray wolf each showed evidence of sexual dimorphism in shape for at least 1 view (p<0.05; Table 5). However, the gray wolf was the only species with significant sexual dimorphism (p<0.002, following Bonferroni adjustment for multiple tests). In the gray wolves, sex was responsible for only a small portion of the total variance (<2.5% in the mandible, <2% in the ventral view), with size accounting for the majority of the shape differences between the sexes. Because variation in size is removed through Generalized 12 Procrustes Analysis, male and female gray wolves were combined. In addition, specimens of unknown sex were added (where available) to species samples. Species Differences.— All species were significantly different from one another in shape for all views (p<0.0004), and all specimens were assigned to the correct species/view by the corresponding assignment test. The jack-knife groupings test correctly classified 100% of the specimens in the lateral and mandibular views, and nearly 99% in the ventral view. The percentages of variance explained by the CVs for each view are summarized in Table 6. In the lateral view, the first 2 CVs explain over 96% of the variance; these CVs are plotted against one another in Figure 3, along with their respective deformation grids. The first CV clearly separates African lions and polar bears (with high scores on CV1) from the remaining species. Morphological changes in transitioning from left to right along CV1 include marked reduction in concavity of the facial profile and reduction in the posterior margin of the sagittal crest. The jugal also expands slightly and moves to a more anterior position relative to the rest of the skull. The second CV in this view separates the 3 hyenas, which have a very prominent sagittal crest and expansive zygomatic region, from the other species, in which those regions are not as strongly emphasized. The hyenas also show a significant reduction in the size of the jugal within the zygomatic arch relative to the rest of the skull and in the area between the infraorbital and lacrimal foramen, along with an anteriorally-directed shift in the postorbital process. In the mandibular view, the first 2 CVs explain almost 80% of the variance across species (Table 6). On CV1 of the corresponding plot (Figure 4), striped hyenas (low scores) and African 13 lions (high scores) are at the extremes, with the remaining species falling near the middle of the axis. The associated deformation grid shows a dorsally directed shift in the ventral margin of the mandibular body from the middle of the tooth row to where the angular process joins the ramus, resulting in a more ventral placement of the angular process relative to the rest of the mandible. This shape change discriminates the striped hyena from the spotted and brown hyenas with the latter having a more ventrally placed angular process relative to the rest of the mandible. There is also an expansion of the dorsal ramus relative to the rest of the mandible, and a broadening of the condyloid and coronoid processes along CV1. On the second CV, the 3 caniforms (i.e., polar bear, gray wolf, African wild dog) form a tight cluster that is separated from the feliforms (i.e., hyenas, lions). Graphical representations of the changes along this CV axis show a reduction in the relative depths of the ramus and the mandibular body below the tooth row, and a marked widening of the mid-dorsal ramus in caniforms compared to feliforms. The tooth row expands anteroposteriorally behind the canine relative to the rest of the mandible. The scale used for the ventral view plot is significantly smaller than that used for the other 2 views, showing that species differences dependent on our choice of landmarks in the ventral view do not discriminate the species from one another as strongly as those in the other views. The first 2 CVs explain nearly 75% of the variance across species (Table 6). The first CV separates species with a relatively rounded skull (low scores) from those with a more streamlined ventral skull morphology (high scores, Figure 5). Shape change is marked by a slight medial compression in the tooth row, contraction of the palatine, and a narrowing and posterior displacement of the zygomatic arches. The area between the external auditory meatus is also shifted posteriorally and compressed medially relative to the rest of the skull. The second CV contrasts the hyenids (low scores) with the polar bear (high scores). Changes along this axis 14 emphasize a narrowing of the zygomatic breadth, relative to the rest of the skull, in non-bone cracking species. The tooth row shows compression similar in direction and scale to that seen in CV1. The area between the external auditory meatus is medially compressed relative to the rest of the skull in the polar bear compared to the hyenids, but there is no posteriorally-directed shift as seen in CV1. Partial Least Squares Analysis.— Results from the PLS analyses are shown in Tables 7 and 8. A lack of independence between shape and the ecological variables is supported by the permutation tests for all 3 views (Table 7). The RV coefficient for each view is high and positive (0.66 for lateral, 0.73 for mandible, and 0.39 for ventral), signifying a strong degree of correlation between the Procrustes coordinates and the ecological index. The first 2 PLS axis pairs explain more than 94% of the total covariance between shape and ecology in every view, with the first alone explaining at least 86% of total variance. Loadings for the ecological variables (Table 8) on the first 2 PLS axis pairs are remarkably similar across views, with killing strategy and hunting party both highly positive and bone consumption having the only negative loading for PLS1. Partial least square axis pair 2 is characterized by a highly positive loading for hunting party, a moderately positive loading for bone consumption, and negative loadings for meat consumption and killing strategy. The relationship between shape and the ecological variables represented by PLS1 is depicted for each view in Figures 6-8. Changes in morphology along the shape PLS1 axis (Figure 6) are associated with a shift from negative to positive ecological scores along the ecology PLS1 axis, where positive ecological scores are associated with an anterodorsal 15 elevation of the rostrum, a reduction in the height of the sagittal crest region relative to the crown of the skull (with the exception of the nuchal crest, which shows dorsally-directed expansion), and dorsal and anteroposterior expansion of the jugal. The non-hyenids also show an anteroposterior expansion of the cranium posterior to the tooth row. Shape change in the mandible (Figure 7) follows a nearly linear relationship with the ecological PLS scores. A higher score on the ecological PLS axis (i.e., species that consume little to no bone) is associated with an increase in mandibular depth anterior to and below the canine and in the angular process, and a reduction in the depth of the mandible below where the tooth row meets the ramus. The ramus/tooth row juncture expands dorsally relative to the rest of the mandible while the posterior margin of the ramus expands posteriorally, and the tooth row expands along the anteroposterior axis relative to the rest of the mandible. The relationship between shape in the ventral view and the ecological variables for PLS1 is depicted in Figure 8. Shape change along PLS1 is similar to ventral CV2 for species, with a medially-directed narrowing of the maxilla, zygomatic arches, and the area between the external auditory meatus, and anteroposterior reduction in the maxilla relative to the rest of the skull. The premaxilla and the area between external auditory meatus expand slightly along the anteroposterior axis relative to the rest of the skull. Anteroposterior expansion from the palate to the back anterior edge of the foramen magnum is emphasized more strongly here than in ventral CV2. Phylogenetic Signal.— 16 The p-values for the permutation tests of the hypothesis of no phylogenetic signal in shape are p<0.0312, p<0.0082, and p<0.0011 for the lateral, mandibular, and ventral views, respectively. Therefore, the null hypothesis is rejected for all 3 views, as all show significant evidence of phylogenetic signal in skull shape. The results of the phylogenetic signal analysis for the ecological variable are shown in Table 9. Bone eating has the lowest K, and the only value less than 1, indicating random or convergent evolution in the acquisition of bone consumption among the hypercarnivores. The K value for hunting group is very close to 1, indicating that evolution of hunting group follows Brownian motion and has some phylogenetic signal. The p-values are significant for killing method across all branch length methods, suggesting non-random phylogenetic signal for killing method. Meat consumption falls marginally short of being significant for all branch length estimations. DISCUSSION There is a high degree of morphological variation in the skulls of the hypercarnivorous members of Order Carnivora, both within and across families. The results suggest that refinement of the classification system put forth by Van Valkenburgh (2007) is needed to adequately capture the ecomorphological forms seen among extant species. Key features, such as the shape of the dorsal curvature of the cranium and the position of the angular process relative to the ventral margin of the mandible, may be useful in further subdividing and discriminating between the hypercarnivorous ecomorphs. Phylogenetic relatedness plays a significant role in the distribution of both morphological and ecological characteristics 17 throughout Carnivora, but this does not depreciate the value of a more refined ecomorphological classification system. Analysis of the mandible resulted in a clear separation of feliform (represented by the African lion and the 3 hyena species) and caniform species (the polar bear, gray wolf, and African wild dog), reflecting the deeper mandibular body and narrower dorsal ramus of members of the feliform lineage. These results compliment the findings of Raia (2004), who found that phylogeny accounted for the greatest portion of variation in mandible shape in a similar study. These differences in shape likely reflect differences in attachment, size, and configuration of the masseter and temporalis muscles, resulting in different areas of focused strain during mastication. Indeed, muscle usage and development has been shown to play a part in the development of bone morphology (Horowitz and Shapiro, 1955). Caniforms and feliforms may differ in the ways they dissipate the stresses associated with biting and food processing, as suggested in stress distribution analyses performed by Tseng and Binder (2010), in which finite element models showed significant differences in mandible strain between spotted hyenas and gray wolves. Identification of distinguishing skull characteristics, such as the relative size of muscle attachment sites and areas important in stress dissipation, is beneficial to paleontological studies, as it offers a starting point from which to make inferences about familial relationships among extinct taxa. Further, such characteristics can be used to identify possible instances of convergent evolution in morphological features. The results suggest that the wolf-like ecomorph may contain a greater range of shape variability across its members than is true for the other 2 hypercarnivorous sub-categories, and a greater degree of shape variability than feliforms. The wide range of skull morphologies seen in the caniforms reflects the wide range of ecological traits observed across the group. In general, 18 caniforms have a narrower and more elongated skull compared to feliforms, but the location on the skull where this manifests, and the degree to which this elongation occurs, varies across species. The gray wolf, with its anteriorally elongated tooth row and rolling brow curvature, represents the average caniform morphology, whereas the African wild dog shows comparatively less anterior expansion in the tooth row and a steeper inclination of the brow. The relatively long skull of the polar bear is achieved very differently, with most of the elongation centered behind the tooth row. Given these marked differences in skull morphology among caniforms, further refinement of the wolf-like ecomorph might be valuable to more adequately describe the diversity among living and extinct taxa. The wolf-like ecomorph might be best split into 3 subcategories based on diet, with 1 group including species with meat-exclusive diets (e.g., the polar bear), another housing species with diets high in tough foods (e.g., the African wild dog and the giant panda), and the final group including those with more omnivorous tendencies (e.g., the gray wolf). One unexpected finding was the repeated grouping of the polar bear and the African lion with respect to skull shape in the lateral and mandibular views. The smoothing of the dorsal curvature and high degree of postorbital elongation of the cranium shared by both species are associated with stalking behavior, as killing strategy discriminated the polar bear and the African lion from all the other species. Both species also have meat-exclusive diets with little to no bone consumption. Thus, features important to the consumption of hard foods, such as the broad zygomatic arches and vaulted forehead exemplified by the spotted hyenas (see Tanner et al., 2008), are absent in both polar bears and African lions. Given the similarities in killing strategy and diet between these species and the associated similarities in cranium shape, the polar bear fits better in the cat-like ecomorph (Van Valkenburgh, 2007) than in the wolf-like ecomorph. 19 Studies with additional hypercarnivorous caniforms would help determine whether the polar bear is unique and deserves to be re-classified as a cat-like ecomorph, or if a sub-categorization scheme as described above would be the better approach. The classification of the African wild dog in the wolf-like ecomorph seems misleading, as this species shares morphological characteristics with both the gray wolf and hyenas. The African wild dog and gray wolf share an anteroposteriorally elongated jugal, dorsoposteriorally slanted orbitals, and a jugal-squamosal suture that extends to the tip of the jugal and forms the ventroposterior base of the orbit. This last characteristic may be unique to Canidae, as it was not seen in any of the other families. The African wild dog and hyenas share a dorsal expansion of the sagittal crest relative to the rest of the skull, a trait that is thought to be important for breaking bones, and nasals that do not extend as far posteriorally as in the other species. The African wild dog displays dorsoanterior inflation in the brow, resulting in a steep incline from the muzzle to the top of the cranium. A similar morphology is seen in the hyenas in association with a caudally elongated frontal sinus, which is thought to aid in stress dissipation during osteophagy (Tanner et al., 2008). The maxilla and zygomatic arches are also broader laterally relative to the skull in the African wild dog and the hyenas than in the other species examined. In contrast to the cranial similarities seen in the African wild dog and the hyenas, the mandible clearly discriminates the two groups, placing the African wild dog in close proximity to the gray wolf. The greater depth of the mandible and the dorsoventrally broader jugal seen in the hyenas, but not in the African wild dog, may suggest differences in the requirements and stresses associated with bone eating between the 2 groups. Reduction in the size of the jugal relative to the rest of the skull as seen in the hyenas, but not in the African wild dog, may be related to stress distribution during bone breaking in the hyenas, as the end of the jugal anterior 20 to the jugal-squamosal suture was found to be a high stress point during biting in finite element analyses by Tanner et al. (2008). The similarities in form and function seen between the hyenas and the African wild dog, and also between the polar bear and African lion, are further evidence that the wolf-like ecomorph may require sub-categorization based on dietary habits. The hyenas are of special interest in that, despite close historical relatedness and ecological similarities, differences in shape discriminate them clearly from one another in all 3 cranial views. Brown and striped hyenas are often seen to be more extreme in shape than the spotted hyena when compared to the other species. This is particularly evident in the ventral view, where brown and striped hyenas differ from spotted hyenas in having anteroposterior expansion of the maxilla and dorsal expansion of the palatine relative to the rest of the skull, and a dorsally-directed reduction in size of the areas posterior to the postglenoid processes. The spotted hyena shows a greater dorsoanterior expansion in the sagittal crest, and a more developed nuchal crest. The jugal also shows a greater contraction relative to the rest of the skull in the spotted hyena than in the other hyenas. These differences in shape may reflect the different dietary habits of the hyenas. Spotted hyenas primarily hunt for food, whereas the majority of meat consumed by the other 2 hyenas is scavenged from carcasses left by other predators, drought, or disease. The scavenging hyenas also compliment their diets with a vast assortment of other foods when vertebrate prey is not available, resulting in a more omnivorous diet compared to spotted hyenas. In addition, differences may be related to the size of the bones being consumed. Spotted hyenas often take down prey considerably larger than themselves, while the other hyenas rely more on carcasses and the occasional small prey item. Kruuk (1975) stated that fecal samples from striped hyenas contain a relatively small amount of large prey items compared to that of spotted hyenas, suggesting differences in the sizes of the bones 21 consumed by each species. Although the differences among the hyenas are not great enough to warrant refinement of the hyena-like ecomorph, they do show that spotted hyenas differ significantly in shape from the scavenging members of Hyaenidae. Phylogeny was expected to play a significant role in the distribution of the ecological traits due to the way in which they were assigned (i.e. the distribution of traits such as scavenging and pack hunting). It was not surprising then that killing strategy, meat consumption, and hunting party all show a phylogenetic signal (K>1). Of these traits, only killing strategy was found to be statistically significant, possibly reflecting the small sample size (i.e., number of species). Evidence of a phylogenetic signal was also expected for skull morphology. Raia (2004) found that phylogeny accounts for a significant amount of shape variation in the carnivoran mandible. He attributed this to the fact that closely related species share a common ancestor and thus a common ancestral shape. While the shape of the cranium may be affected by more functional requirements than the mandible, similarities due to common ancestry will still be noticeable throughout Carnivora. Identifying morphological characteristics shared by taxa with similar behavioral ecologies is an important step in furthering our ability to make inferences about the behaviors of both living and extinct species. Moreover, variation in the shape of the skull, and its dietary and ecological implications, are key to studies of carnivoran evolution. The ability to look at shape difference throughout the skull as a complete entity is vital to furthering our understanding of these differences, and will permit us to further study the interactions and associations among the morphological features that make up the feeding apparatus. Future ecomorphological analyses would benefit greatly from a broader usage of geometric morphometric techniques, as they offer the ability to visualize both small and sweeping changes in shape across the entire skull when 22 used in conjunction with ecological assessments. Through our analyses, we were able to identify several morphological features that can be used to discriminate hypercarnivorous species from one another. While feature such as a greater steepness of the brow and a prominent sagittal crest (features thought to be associated with the bone eating behavior seen in the hyenids) have been described in the past, the differences in the relative shape of the jugal (seen and the position of the angular process relative to the ventral margin of the mandible have not been previously identified. The results presented here suggest that the hypercarnivore classification system described by Van Valkenburgh (2007) may be in need of revision. While the system put forth by Van Valkenburgh has been a useful tool for ecomorphologists looking to assess the possible ecological behaviors of extinct species, our analyses show that the categorization method used may be too broad to accurately make such inferences. Our analyses show this to be especially true for the wolf-like ecomorph, which Van Valkenburgh describes as being not as extreme in skull and dental modifications as the other 2 ecomorphs. A classification scheme which subdivides the wolf-like ecomorph into several distinct categories would be better suited for use in future ecomorphological studies. A possible classification scheme would include a generalized wolf-like ecomorph, similar to the gray wolf, and a durophagious category with species like the African wild dog. If further analyses find that the polar bear is not unique in its similarities to the cat-like ecomorph, then a third category bridging the gap between the cat-like and wolf-like ecomorph may also be needed. A more encompassing analysis (e.g., more taxa, 3D visualization, greater ecological classification) would be beneficial to determining the extent to which such categories should be broken down, and what morphological features exemplify each category. 23 Table 1 - Chart from Van Valkenburgh 2007 showing the occurrence of ecomorphs throughout the evolution of the carnivoran families, with emphasis on terrestrial species larger than 7kg in mass. A “+” indicates that at least one species in the fossil record is known to have displayed a particular ecomorph. “?” indicates that the existence of a species with a particular ecomorph is unknown given the limited fossil record. 24 Table 2 – Descriptions of the common behavioral traits assigned to prey acquisition and consumption for each species used in the study following a literature review. Characteristics assigned to species are taken as common tendencies and do not reflect behaviors that occur infrequently. Species Meat Bone Hunting Party Spotted Hyena Mostly About 10% Alone or small groups Some Forage and feed alone Striped Hyena Brown Hyena Polar Bear As available, supplemented with other foods As available, supplemented with other foods Almost all Some Little Forage alone, sometimes gather at large carcasses and feed in turn Alone or in groups depending on food abundance African Wild Dog Mostly About 10% Pack Gray Wolf Mostly Smaller bones Pack African Lion Almost all Smaller bones Usually hunt in groups Strategy Hunters and scavengers, prey is often large, large prey ripped open, small prey killed with bite to head/neck Sources MacNulty et al. 2007, OwenSmith and Mills 2008, Van Valkenburgh 1996 Primarily scavenging, very little killing of prey and then only of small animals Leakey et al. 1999, Wagner 2006 Primarily scavenging, very little killing of prey and then only of small animals Mills 1982, Wiesel 2006 Wait and stalk, kill with brute force, no specific killing posture, large prey but frequently small compared to them Stalk and chase, smaller prey are mobbed, larger prey slashed with teeth until shocked/exhausted, prey often eaten alive, prey are generally as big or bigger than them Pursue and/or harass followed by multiple bites, mainly large prey opportunistic scavengers when possible (up to 40% of diet), most prey is large, stalk or pursue and pounce, killing bite and twist? 25 DeMaster and Stirling 1981, Sacco and Van Valkenburgh 2004 Estes and Goddard 1967, MacNulty et al. 2007, OwenSmith and Mills 2008, Van Valkenburgh 1996 MacNulty et al. 2007, Mech 1974, Stahler et al. 2006 Haas et al. 2005, MacNulty et al. 2007, Owen-Smith and Mills 2008, Tsukahara 1993, Van Valkenburgh 1996 Table 3 – Index of ecological categories and the values assigned to them for use in partial least squares analysis. Values for ecomorph were ordered arbitrarily. Value 1 Meat Consumption As available, omnivorous diet Bone Consumption Hunting Party Killing Strategy Little to none Solitary Primarily scavenging Chase, knock down, tear apart 2 Most of diet Smaller bones only, little portion of diet Occasionally in groups (depending on food abundance) 3 Nearly exclusive easily-broken bones, significant part of diet Alone or small groups Chase, bite/exhaust, tear apart 4 --- Any that can be broken, considerable part of diet Packs or large groups Stalk and pounce, killing bite 5 --- --- --- Stalk, kill with brute force 26 Table 4 – Values assigned to each species following the ecological index in Table 3. Assignment of values was performed by comparing the descriptive summaries from Table 2 to the categories described in Table 3. Species Meat Bone Spotted Striped Brown Polar Wild Dog Wolf Lion 2 1 1 3 4 3 3 1 Hunting Party 3 1 1 2 2 4 4 3 2 3 2 2 4 4 3 4 27 Strategy 2 1 1 5 Table 5 – Results from the Goodall’s F test with permutation for signs of sexual dimorphism in shape, and the subsequent Mancova for shape with size as a covariate that were significant prior to Bonferroni adjustment (p<0.002). % Variance % Variance Total % explained explained variance by size by sex explained 0.0172 26.19 9.9548 36.1448 0.0352 0.0432 23.297 5.3505 28.6475 African Lion – Ventral 0.0476 0.6312 11.871 2.3934 14.2644 Gray Wolf – Mandible 0.0016 0.014 8.645 2.4659 11.109 Gray Wolf – Ventral 0.0004 0.0792 5.778 1.7393 7.5173 Goodall’s Mancova p-value p-value Polar Bear – Mandible 0.0116 African Lion – Mandible Species – View 28 Table 6 – Canonical variates (CVs) obtained for all 3 views with the percent of variance explained by each CV and cumulative percentage of explained variance for that view. View CV Eigenvalues % Variance Cumulative % Lateral 1 1596.862879 90.155 90.155 2 120.094379 6.78 96.935 3 31.07215079 1.754 98.69 4 12.56631796 0.709 99.399 5 7.47116971 0.422 99.821 6 3.17229624 0.179 100 1 2269.458657 57.375 57.375 2 862.7065408 21.81 79.186 3 561.0455231 14.184 93.37 4 160.1684887 4.049 97.419 5 80.52965515 2.036 99.455 6 21.55710116 0.545 100 1 47.09877689 49.749 49.749 2 23.77994285 25.118 74.867 3 16.59380728 17.527 92.394 4 4.51629731 4.77 97.165 5 2.07546399 2.192 99.357 6 0.60870517 0.643 100 Mandible Ventral 29 Table 7 – Results from partial least squares analyses examining covariation between skull morphology and ecological variables (Table 4). The RV coefficient for a given view is inserted below the name of the corresponding view in the first column. Independence % Total Correlation Test P-value Covariance (permutation) View Variable Singular Value Lateral PLS1 0.08538319 <.0001 88.946 0.93636 RV PLS2 0.02783738 <.0001 9.454 0.74652 0.6595 PLS3 0.0113086 <.0001 1.56 0.68632 PLS4 0.00179605 <.0001 0.039 0.61973 Mandible PLS1 0.0666852 <.0001 91.77 0.93929 RV PLS2 0.01810571 <.0001 6.765 0.80229 0.7309 PLS3 0.00829509 <.0001 1.42 0.63083 PLS4 0.00147778 <.0001 0.045 0.52652 Ventral PLS1 0.06819104 <.0001 85.62 0.80991 RV PLS2 0.02191763 <.0001 8.845 0.83781 0.3900 PLS3 0.01720183 <.0001 5.448 0.60626 PLS4 0.00216774 <.0001 0.087 0.63211 30 Table 8 – Weighting coefficients assigned to each ecological variable for the first 2 pairs of partial least square axes for each of the 3 views following partial least squares analysis. View Variable PLS1 PLS2 Lateral Meat 0.34883278 -0.34819878 Bone -0.31468029 0.1965845 Hunting Group 0.58584705 0.80563216 Kill Strategy 0.66035993 -0.43711436 Meat 0.36140411 -0.24619879 Bone -0.29723917 0.35997854 Hunting Group 0.56258882 0.81376489 Kill Strategy 0.6815642 -0.38417225 Meat 0.27946663 -0.2647549 Bone -0.38624738 0.32677728 Hunting Group 0.61971087 0.78340854 Kill Strategy 0.62343388 -0.45759425 Mandible Ventral 31 Table 9 – Blomberg’s (2003) K statistic and the p-values for tests of phylogenetic independent contrast variance under each of the 4 branch length estimation methods tested for all ecological variables, and for the consensus tree obtained from the 4 single trees. 1 = all branch lengths set to 1.0, G = Grafen (1989), N = Nee (cited in Purvis 1995), and P = Pagel (1992). P-values <0.05 show signs of non-random phylogenetic signal. Variable K PIC Variance p-value (1) PIC Variance p-value (G) PIC Variance p-value (N) PIC Variance p-value (P) PIC Variance pvalue (Consensus) Meat 1.495 0.0755 0.0605 0.0565 0.0555 0.057 Bone 0.695 0.584 0.5505 0.536 0.562 0.5475 Hunting group 1.055 0.096 0.0805 0.092 0.1005 0.0995 Killing Strategy 1.647 0.0365 0.037 0.028 0.028 0.0365 32 Figure 1 - - Lateral, mandibular, and ventral views of a Crocuta crocuta skull, showing locations of landmarks (red, numbered dots, see Table 2), and semi-landmarks (red, unnumbered triangles). For interpretation of the references to color in this and all other figures, the reader is referred to the electronic version of this thesis. 33 Figure 2 - Phylogenetic tree of the 7 Carnivora taxa included in this study. Relationships are based on Eizirik et al. 2010. 34 Figure 3 - Plot of lateral view CV1 vs. CV2 for species, with deformation grids depicting shape change along CV1 (bottom right) and CV2 (top left). The legend (bottom left) lists the symbol associated with each species. 35 Figure 4 - Plot of mandibular view CV1 vs. CV2 for species, along with its deformation grids and legend as described in Figure 3. 36 Figure 5 - Plot of ventral view CV1 vs. CV2 for species, along with its deformation grids and legend as described in Figure 3. Note the axis ranges are significantly less than that seen in the other 2 views. 37 Figure 6 – Plot of partial least squares for shape axis 1 vs. ecology axis 1 for the lateral view. A plot of PLS1 vs. PLS2 for the ecological traits showing the directionality and relative strength of the trait coefficients is inserted to the right of the plot. Below the plot is a deformation grid showing the shape change associated with left to right movement along shape PLS1. 38 Figure 7 – Plot of partial least squares for shape axis 1 vs. ecology axis 1 for the mandibular view. See Figure 6 for further explanation. 39 Figure 8 – Plot of partial least squares for shape axis 1 vs. ecology axis 1 for the ventral view. See Figure 6 for further explanation. 40 APPENDICES 41 APPENDIX A List of included museum specimens 42 Table 10 - List of specimens used in this study. Abbreviations within the Museum column represent where the specimen is located. (MSU = Michigan State University, East Landing, MI, TAU = Tel Aviv University, London, England, BMNH = British Museum of Natural History London, England, FMNH = Field Museum of Natural History, Chicago IL, NMK = National Museums of Kenya, Africa, RCSOM = Royal College of Surgeons, London, England). Specimens whose sex could not be determined were marked “?” in the Sex column. Lateral, ventral, and mandible refer to the digitization views used. Those specimens marked "Y" were included in the corresponding view analyses, whereas that marked with a "-" were not. Museum ? ? ? ? MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU Number 225 VGS 486 ECO 799 HUM 897 BFT 35852 35853 35854 35855 35856 35857 35858 35859 36008 36009 36011 36074 36077 36078 36079 36080 36081 36082 36083 36084 36094 36156 36160 36161 36162 36163 Species Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta 43 Sex Lateral Ventral Mandible M Y F Y M Y M Y M Y Y Y M Y Y Y M Y Y Y M Y Y Y F Y Y Y ? Y Y Y ? Y Y Y F Y Y Y F Y Y Y ? Y Y Y F Y Y Y F Y Y Y F Y Y Y M Y Y Y M Y Y Y F Y Y Y F Y Y Y F Y Y Y F Y Y Y M Y Y Y F Y Y Y F Y Y Y F Y Y Y ? Y Y Y M Y Y Y M Y Y Y Table 10 (cont’d) MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU TAU TAU TAU BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH 36164 36165 36167 36168 36550 36551 36552 36553 36556 36558 36566 36567 36568 36569 36570 36571 36576 36580 36581 2208 8005 7026 0.5.12.1 1096 1495 1937.10.30 1938.10.18.48 1990.389 20.10.27.1 20.10.27.2 23.1.1.78 23.1.1.79 23.1.1.80 23.3.4.9 24.10.5.5 26.10.8.72 26.10.8.73 Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Crocuta crocuta Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena 44 ? F ? M F F F F F F ? F ? F F F ? ? F ? F M ? F ? F ? ? M F M F F M ? M F Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y - Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y - Table 10 (cont’d) BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH 27.2.14.27 31.1.2.10 34.11.28.10 34.11.28.11 34.11.28.12 34.11.28.13 34.11.28.16 34.11.28.18 34.11.28.3 34.11.28.4 34.11.28.5 34.11.28.6 34.11.28.8 34.8.4.7 35.1.1.1 35.1.1.2 38.8.4.6 39.439 39.44 44.2.28 47.36 5.5.28.2 51.8.25.1 52.1483 56.5.6.50 58.209 58.6.24.125 6.5.4.3 8.7.24.12 85.6.13.1 85.8.1.50 92.2.8.2 HhNo#1 HhNo#2 15.3.6.18 1938.6.28.4 2.11.22.5 Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena ? ? F M M M M F M F M F M M ? ? ? M F ? F ? ? F ? ? ? M ? ? M M ? ? M ? M 45 Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y - Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Table 10 (cont’d) BMNH BMNH BMNH BMNH BMNH FMNH FMNH FMNH FMNH FMNH FMNH FMNH FMNH MSU MSU MSU NMK NMK NMK NMK RCSOM RCSOM RCSOM RCSOM RCSOM TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU 2.11.4.2 24.10.321 24.10-5.6 34.8.4.6 93.12.1.1. 103991 107342 140215 140216 140218 140219 140220 103992 11143 13003 36395 3474 4628 8297 6297 137.21 137.3 137.31 137.33 144.44 10236 10616 10617 10683 11099 11130 11248 11249 11515 11533 11687 11821 Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena ? ? ? M ? M ? ? F F ? ? M ? ? M M ? ? ? ? ? ? ? ? F F M M M F F M M ? M F 46 Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Table 10 (cont’d) TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU 11846 11945 12128 2019 22 2484 2536 2714 276 2814 3201 3316 3597 4035 4376 4746 5106 5127 5379 594 6140 6202 6444 6510 6640 6677 6804 6895 7 7119 7216 7217 7238 7256 7335 7336 7455 Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena F M ? F M M F M M ? M F F M F F M F F F F M F M ? F M M M F M ? F M F F M 47 Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Table 10 (cont’d) TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU TAU BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH BMNH FMNH 7480 7502 7618 7644 7672 7737 7813 7839 7898 7962 8294 8295 8666 9010 9160 9418 9423 9739 9743 9811 9930 8037 9715 79.1631 26.12.7.330 26.12.7.331 26.12.7.332 26.12.7.333 35.9.1.284 35.9.1.285 35.9.1.286 35.9.1.287 35.9.1.288 46.7.2.7 53.3.11.1 66.5.2.1 34585 Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Hyaena hyaena Parahyaena brunnea Parahyaena brunnea Parahyaena brunnea Parahyaena brunnea Parahyaena brunnea Parahyaena brunnea Parahyaena brunnea Parahyaena brunnea Parahyaena brunnea Parahyaena brunnea Parahyaena brunnea Parahyaena brunnea Parahyaena brunnea Parahyaena brunnea 48 F F ? F F M ? F ? M F M M M M M M F F M F M M ? F ? ? ? ? F ? M F ? ? ? F Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Table 10 (cont’d) FMNH MSU NMNH NMNH MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU RCSOM TAU TAU TAU TAU TAU TAU BMNH BMNH BMNH BMNH MSU MSU MSU MSU 34586 33836 296134 429178 3949 4251 4851 11240 11241 11242 11674 11675 12236 12392 16791 20127 20954 21884 24411 29954 29988 28292 114.2 21 2552 2810 6614 7308 7638 31.1.2.3 31.1.3.2 35.3.14.3 43.64 14954 20126 24405 36073 Parahyaena brunnea Parahyaena brunnea Parahyaena brunnea Parahyaena brunnea Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo 49 F M F F M M M M M M M M M M M M M M M M M M M M M M M M M F F F F F F F F Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Table 10 (cont’d) RCSOM TAU TAU TAU BMNH BMNH ? MSU MSU NMK NMK NMK TAU BMNH BMNH BMNH BMNH NMK NMK NMK NMK NMK NMK NMK NMK NMK NMK NMK TAU TAU TAU TAU TAU TAU TAU TAU TAU 114.91 1932 2553 3916 14.4.12.190 36.3.14.24 K5482 KayNo# 8046 2523 4940 4948 No# 58.213 58.226 75.139 67.4.12.188 2632 2636 2639 2640 2643 3404 7425 7436 7439 7516 7889 4003 4005 4006 4451 4767 5235 5437 5575 7025 Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Panthera leo Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus Lycaon pictus F F F F ? ? ? ? ? ? ? ? ? ? ? ? ? F ? F F ? ? M ? ? M ? M M F F M M M F F 50 Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Table 10 (cont’d) MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU ? MSU MSU MSU MSU MSU MSU 11673 20963 22148 27987 36573 3696 4686 9322 9323 9324 9326 9327 9328 9330 9332 9335 12760 2132 23998 24317 24432 9325 9329 9333 9334 16588 27845 33107 9312 9331 DSCF0320 10596 10663 24321 35868 35884 36213 Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Ursus maritimus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus 51 M M M M M M M M M M M M M M M M F F F F F F F F F ? ? ? M ? ? M M M M M F Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Table 10 (cont’d) MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU 36243 36248 36251 36385 36387 36393 36394 36418 36420 36421 36422 36423 36424 36427 36428 36430 36433 36442 36445 36446 36447 36448 36449 36450 36451 36452 36453 36515 36516 36518 36519 36520 36521 36522 36523 36524 36527 Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus M F F F M F F F M F M F F F F F M F F M M M F F M M M F M M M M F M M M F 52 Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Table 10 (cont’d) MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU 36531 36533 36534 36535 36536 36538 36539 36541 36542 37149 37150 37151 37152 37153 37154 37155 37156 37157 37158 37159 37160 37161 37162 37163 37164 37165 37166 37167 37168 37169 37170 37171 37172 37173 37174 37175 37176 Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus F F M M F M M M F F M M M M M F M M F M M M M M F M M F M M F F M F M M F 53 Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y - Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Table 10 (cont’d) MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU MSU 37177 37178 37179 37180 37181 37182 37183 37184 37185 37186 37187 37188 37189 37190 37191 37192 37193 37195 9670 Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus Canis lupus F F F M F M F M F F F F M F M F M F M 54 Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y APPENDIX B Locations of landmarks and semi-landmarks 55 Table 11 - Locations of landmarks and semi-landmarks for each of the three views. Lateral Landmarks 1 Anterior point of I3 2 Anterior point of canine 3 Posterior point of canine 4 Anterior point of the infraorbital foramen 5 Upper-most point of the lacrimal foramen 6 Tip of the post-orbital process 7 Dorsal edge of the jugal-squamosal suture 8 Ventral edge of the jugal-squamosal suture 9 Ventral tip of the jugal 10 Anterior-most point along the curve of the pterygoid 11 Upper-most point on the external auditory meatus 12 Upper-most point on the occipital condyle 13 Posterior-most point on the nuchal crest 14 Anterior-most point on the nasal-premaxilla suture 15, 16 10mm scale 17-48 Semi-landmarks evenly space from 14 to 13 Mandible Landmarks 1 Anterior point of the I3-dentary boundary 2 Anterior point of the canine-dentary boundary 3 Posterior point of the canine-dentary boundary 4 Upper edge of the mental foramen 5 Dorsal apex of the curve on the coronoid process 6 Posterior-most point of the coronoid process 7 Anterior edge of the mandibular condyle 8 Posterior-most point of the mandibular condyle 9 Dorsal tip of the articular process 10 Posterior point of the tooth row 11 Anterior point of the I1-dentary boundary 12, 13 10mm scale 14-45 Semi-landmarks evenly spaced from 11 to 9 46-56 Semi-landmarks evenly spaced from 9 to 8 57-72 Semi-landmarks evenly spaced from 7 to 6 73-88 Semi-landmarks evenly spaced from 5 to 10 56 Table 11 (cont’d) Ventral Landmarks 1 Juncture between incisors on the premaxilla 2, 5 Intersection of premaxilla-maxilla suture and the medial edge of the canine 3, 4 Posterior point of the incisive foramen 6 Posterior point of the premaxilla-maxilla suture on the palate 7 Maxilla-palatine midline suture 8, 9 Medial curvature of the suture between the maxilla and the palate 10 Posterior-most edge of midline suture between the right and left palatine 11, 12 Medial edge of the maxilla-jugal suture 13, 14 Posterior-most edge along the jugal-squamosal suture 15 Anterior point on the foramen magnum 16, 17 Maxilla-palatine suture at the posterior edge of the palate 18, 19 Posterior tip of pterygoid 20, 21 Center of jugular foramen 22, 23 Medial edge of the glenoid process 24, 25 Posterior edge of P2 26, 27 Anterior edge of the external auditory meatus 28, 29 10mm scale 57 WORK CITED 58 WORKS CITED Blomberg, S. P., T. Garland, Jr., and A. R. Ives. 2003. Testing for phylogenetic signal in comparative data: behavioral traits are more labile. Evolution. 57 (4): 717-745. Christiansen, P. and J. S. Adolfssen. 2005. Bite forces, canine strength and skull allometry in carnivores (Mammalia, Carnivora). Journal of the Zoological Society of London. 266: 133-151. Christiansen, P. and S. Wroe. 2007. Bite forces and evolutionary adaptations to feeding ecology in carnivores. Ecology. 88 (2): 347-358. DeMaster, D. P. and I. Stirling. 1981. Ursus maritimus. Mammalian Species. 145: 1-7. Eizirik, E., W. J. Murphy, K.-P. Koepfli, W. E. Johnson, J. W. Dragoo, R. K. Wayne, and S. J. O’Brien. 2010. Pattern and timing of diversification of the mammalian order Carnivora inferred from multiple gene sequences. Molecular Phylogenetics and Evolution. 56: 49-63. Estes, R. D. and J. Goddard. 1967. Prey selection and hunting behavior of the African wild dog. The Journal of Wildlife Management. 31 (1): 52-70. Ferretti, M. P. 2007. Evolution of bone-cracking adaptations in hyaenids (Mammalia, Carnivora). Swiss Journal of Geosciences. 100: 41-52. 59 Goswami, A. and P. D. Polly. 2010. The influence of modularity on cranial morphological disparity in Carnivora and Primates (Mammalia). PLoS ONE. 5(3): e9517. doi:10.1371/journal.pone.0009517. Grafen, A. 1989. The phylogenetic regression. Philosophical Transactions of the Royal Society of London B 326:119-157. Haas, S. K., V. Hayssen, and P. R. Krausman. 2005. Panthera leo. Mammalian Species. 762: 1-11. Holliday, J. A. and S. J. Steppan. 2004. Evolution of hypercarnivory: the effect of specialization on morphological and taxonomic diversity. Paleobiology. 30(1): 108-128. Horowitz, S. L. and H. H. Shapiro. 1955. Modification of skull and jaw architecture following removal of the masseter muscle in the rat. American Journal of Physical Anthropology. 13(2): 301-308. Kembel, S. W. 2010. An introduction to the picante package. Last accessed March 16, 2011. http://picante.r-forge.r-project.org/picante-intro.pdf Kembel, S. W., P.D. Cowan, M.R. Helmus, W.K. Cornwell, H. Morlon, D.D. Ackerly, S.P. Blomberg, and C.O. Webb. 2010. Picante: R tools for integrating phylogenies and ecology. Bioinformatics 26:1463-1464. Klingenberg, C. P. 2011. MorphoJ: an integrated software package for geometric morphometrics. Molecular Ecology Resources 11: 353-357. doi: 10.1111/j.1755-0998.2010.02924.x Koehler, C. E. and P. R. K. Richardson. 1990. Proteles cristatus. Mammalian Species. 363: 1-6. Koepfli, K.-P., S. M. Jenks, E. Eizirik, T. Zahirpour, B. Van Valkenburgh, and R. K. Wayne. 2006. Molecular systematic of the Hyaenidae: relationships of a relictual lineage resolved by a molecular supermatrix. Molecular Phylogenetics and Evolution. 38: 603-620. 60 Kruuk, H. 1975. Hyaena. Oxford University Press. Leakey, L. N., S. A. H. Milledge, S. M. Leakey, J. Edung, P. Haynes, D. K. Kiptoo, and A. McGeorge. 1999. Diet of the striped hyena in northern Kenya. African Journal of Ecology. 37: 314-326. MacNulty, D. R., L. D. Mech, and D. W. Smith. 2007. A proposed ethogram of large-carnivore predatory behavior, exemplified by the wolf. Journal of Mammalogy. 88 (3): 595-605. Maddison, W.P. & D.R. Maddison. 2010. Mesquite: A modular system for evolutionary analysis. Version 2.74. http://mesquiteproject.org Mech, L. D. 1974. Canis lupus. Mammalian Species. 37: 1-6. Midford, P. E., T. Garland Jr., and W. P. Maddison. 2010. PDAP Package of Mesquite. Version 1.15. Mills, M. G. L. 1982. Hyaena brunnea. Mammalian Species. 194: 1-5. Owen-Smith, N. and M. G. L. Mills. 2008. Predator-prey size relationships in an African large mammal food web. Journal of Animal Ecology. 77: 173-183. Pagel, M. D. 1992. A method for the analysis of comparative data. Journal of Theoretical Biology 156:431-442. Purvis, A. 1995. A composite estimate of primate phylogeny. Philosophical Transactions of the Royal Society of London B 348:405-421. Raia, P. 2004. Morphological correlates of tough food consumption in large land carnivores. Italian Journal of Zoology. 71: 45-50. 61 Renaud, S., J-C. Auffray, and S. de la Porte. 2010. Epigenetic effects on the mouse mandible: common features and discrepancies in remodeling due to muscular dystrophy and response to food consistency. BMC Evolutionary Biology. 10(28). Sacco, T. and B. Van Valkenburgh. 2004. Ecomorphological indicators of feeding behaviour in the bears (Carnivora: Ursidae). Journal of the Zoological Society of London. 263: 41-54. Stahler, D. R., D. W. Smith, and D. S. Guernsey. 2006. Foraging and feeding ecology of the gray wolf (Canis lupus): lessons from Yellowstone National Park, Wyoming, USA. Journal of Nutrition. 136 (supplement): 1923S-1926S. Tanner, J. B., E. R. Dumont, S. T. Sakai, B. L. Lundrigan, and K. E. Holekamp. 2008. Of arcs and vaults: the biomechanics of bone-cracking in spotted hyenas (Crocuta crocuta). Biological Journal of the Linnean Society. 95: 246-255. Tseng, Z. J. and W. J. Binder. 2010. Mandibular biomechanics of Crocuta crocuta, Canis lupus, and the late Miocene Dinocrocuta gigantea (Carnivora, Mammalia). Zoological Journal of the Linnean Society. 158: 683-696. Tsukahara, T. 1993. Lions eat chimpanzees: the first evidence of predation by lions on wild chimpanzees. American Journal of Primatology. 29: 1-11. Van Valkenburgh, B. 1996. Feeding behavior in free-ranging, large African carnivores. Journal of Mammalogy. 77 (1): 240-254. Van Valkenburgh, B. 2007. Déjà vu: the evolution of feeding morphologies in the Carnivora. Integrative and Comparative Biology. 47(1): 147-163. Wagner, A. P. 2006. Behavioral ecology of the striped hyena (Hyaena hyaena). Doctoral dissertation at Montana State University. Werdelin, L. and N. Solounias. 1996 The evolutionary history of hyaenas in Europe and Western Asia during the Miocene. In: Bernor, R.L. et al. The evolution of western Eurasian Neogene mammal faunas, Columbia University Press, New York, 290–306. 62 Wiesel, I. 2006. Predatory and foraging behaviour of brown hyenas (Parahyaena brunnea (Thunberg, 1820)) at cape fur seal (Arctocephalus pusillus pusillus Schreber, 1776) Colonies. Doctoral dissertation at the University of Hamburg, Germany. Wroe, S., C. McHenry, and J. Thomason. 2005. Bite club: comparative bite forces in big biting mammals and the prediction of predatory behavior in fossil taxa. Proceedings of the Royal Society of Biological Sciences. 272: 619-625. Zelditch, M. L., D. L. Swiderski, H. D. Sheets, and W. L. Fink. 2004. Geometric morphometrics for biologists: a primer. Elsevier Academic Press. 63